Linking troubles: ideas requested ;-)

This forum deals with any kind of routing computation whether it is simple A:B-routing, calculation of isochrones, simple matrix computation or nearest search.
Post Reply
User avatar
Bernd Welter
Site Admin
Posts: 2695
Joined: Mon Apr 14, 2014 10:28 am
Contact:

Linking troubles: ideas requested ;-)

Post by Bernd Welter »

Hello together,

today I've been asked for a specific solution for the following situation - see attached image:
  • customer wants to compute time of arrival based on current GPS position and next destination coordinates
  • due to GPS deviation a vehicles on board unit provides the blue location as the current position - we also know that the vehicles heading is 181 (dotted line)
  • what happens is that the vehicle is matched into a low level dead end street (purple icon)
  • what we want to have is the vehicle being matched to the highway in direction 181 (red circle)
red circle = desired match location<br />purple icon = false match<br />blue icon = inaccurate position of GPS
red circle = desired match location
purple icon = false match
blue icon = inaccurate position of GPS
linking troubles.png (20.45 KiB) Viewed 5864 times
We already played with the combination of standard properties coordinates, linking type and the non-standard heading (enabled in the profile, provided in the request) and this resolved another big issue but in this case we need a different approach.

Besides current (coordinates, heading) we also have access to a history of (coordinates, heading)... is there anybody who also tried to implement such a plot based on xRoute (and maybe xMapMatch)?

I need an ignition spark ;-)

Best regards Bernd
Joost
Posts: 310
Joined: Fri Apr 25, 2014 1:46 pm

Re: Linking troubles: ideas requested ;-)

Post by Joost »

The proper way would be to feed the last few points of the track into xMapmatch and do a global matching. If this is doable and how big the trace needs to be depends on the quality of the signal.

Off coarse there is also the "shadey" way: install an extracted instead of detailed map. If there are no small roads in the map, the xRoute cannot link to them :twisted: .
User avatar
Bernd Welter
Site Admin
Posts: 2695
Joined: Mon Apr 14, 2014 10:28 am
Contact:

Re: Linking troubles: ideas requested ;-)

Post by Bernd Welter »

Hello Joost,

I think we have to evaluate the usage of xMapMatch because under some special conditions it is definetly necessary to link the GPS to the low level street - if the recent coordinates seem to belong to that kind of street after the map matching.

Using a poor map with high level streets is definetly not a valid approach, e.g. because the destinations of the users could also belong to low level streets and we need a precise routing.

Best regards and one more time "thanks for the input",

Bernd
Post Reply